An efficient FMO-based error-resilient scheme for video surveillance coding

Abstract

The encoded video bitstreams have to suffer from quality degradation when transmitted over unreliable channels. In this paper, a novel flexible macroblock ordering (FMO)-based error-resilient scheme is proposed to cope with this problem, which is aimed at the video surveillance scenarios. This scheme could not only meet the need of prioritized transmission environment, but also take advantage of FMO's inherent error-resilient function. A motion region extraction algorithm is first advised and utilized, and each motion region's error sensitivity is then estimated based on both pictures' features and transmission condition. According to the error sensitivity value, the optimal FMO encoding mode for each motion region is selected, which enables the error-resilient intensity to be more adaptive. The intra refresh enhancement is introduced at the same time to further improve the error-resilient performance of the proposed FMO structure. To reduce the complexity, most information for calculations is derived from the encoding process. Experiments demonstrate that the proposed scheme performs better than the conventional schemes, and improves the video quality efficiently.